Student : RT Masimba
About the student
I am an Honours student passionate about Artificial Intelligence and its real-world applications. I enjoy building intelligent systems and exploring innovative solutions through research and development. With experience in AI-driven projects and software development, I am always eager to showcase my work and contribute to impactful technological advancements.
About the Project
Breast cancer is one of the leading global health challenges, and early detection is vital for successful treatment. Ultrasound is used as it is safe, affordable, and accessible, particularly in resource-limited settings. However, interpretation of ultrasound scans can be difficult due to reliance on expert judgment. Artificial Intelligence offers potential support, but most systems are “black boxes,” limiting clinical trust. The BreastEcho-XAI project tackles this challenge by developing an interpretable, feature-based classification system for breast ultrasound images. Instead of relying on opaque deep learning models, it extracts transparent features such as shape (area, compactness, circularity), intensity (mean, variance, skewness), and texture (GLCM and LBP descriptors). These features are processed by simple, explainable models including K-Nearest Neighbors, Decision Trees, and Support Vector Machines. A Python-based demo tool was created to demonstrate clinical usability. It allows users to upload ultrasound images, apply preprocessing and feature extraction, and view predictions along with explanations. Results on the BUSI dataset show that transparent models can achieve reliable classification performance while remaining interpretable. This project emphasizes the importance of Explainable AI (XAI) in healthcare, aiming to build more trustworthy, accessible, and clinically relevant diagnostic tools.
